ABSTRACT
The long - term pavement performance of flexible pavements depends on a variety of generally known factors such as mix design, layer thickness, weather conditions and traffic loading. Common pavement performance models are usually based on regression approaches leading to deterministic average performance functions or computed discrete condition-based Markov transition probabilities in discrete or continuous time. However, these approaches have a number of shortcomings that are addressed shortly in the paper. Instead of using these approaches a new stochastic continuous time and state space process is introduced allowing a much more accurate description of any ageing system with continuous condition development and failure/condition distribution over time. In addition, the presented approach allows a survival analysis taking into account already failed/replaced sections, providing a condition based remaining service live of all surviving sections. Furthermore, the calculation of stochastic condition development with measures together with the resulting costs are provided allowing accurate calculations both on project and network level. Finally, the results of an application of this approach to real data from condition surveys from LTPP (USA) are provided.
